- first cut a Go parser in Go

SVN=126242

doc/go_lang.txt

@@ -4,7 +4,7 @@ The Go Programming Language (DRAFT)
 Robert Griesemer, Rob Pike, Ken Thompson
 
 ----
-(July 3, 2008)
+(July 7, 2008)
 
 This document is a semi-formal specification/proposal for a new
 systems programming language.  The document is under active
@@ -314,10 +314,11 @@ Reserved words
   break         fallthrough       import            return
   case          false             interface         select
   const         for               map               struct
-  continue      func              new               switch
-  default       go                nil               true
-  else          goto              package           type
-  export        if                range             var
+  chan          func              new               switch
+  continue      go                nil               true
+  default       goto              package           type
+  else          if                range             var
+  export
 
 
 TODO: "len" is currently also a reserved word - it shouldn't be.

usr/gri/src/parser.go

@@ -0,0 +1,705 @@
+// Copyright 2009 The Go Authors.  All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package Parser
+
+import Scanner "scanner"
+
+
+export Parser
+type Parser struct {
+	verbose bool;
+	S *Scanner.Scanner;
+	tok int;  // one token look-ahead
+	beg, end int;  // token position
+};
+
+
+func (P *Parser) Next() {
+	P.tok, P.beg, P.end = P.S.Scan()
+}
+
+
+func (P *Parser) Open(S *Scanner.Scanner, verbose bool) {
+	P.verbose = verbose;
+	P.S = S;
+	P.Next();
+}
+
+
+func (P *Parser) Error(msg string) {
+	print "error: ", msg, "\n";
+}
+
+
+func (P *Parser) Trace(msg string) {
+	if P.verbose {
+		print msg, "\n";
+	}
+}
+
+
+func (P *Parser) Expect(tok int) {
+	if tok != P.tok {
+		P.Error("expected `" + Scanner.TokenName(tok) + "`, found `" + Scanner.TokenName(P.tok) + "`");
+	}
+	P.Next();  // make progress in any case
+}
+
+
+func (P *Parser) ParseType();
+func (P *Parser) ParseExpression();
+
+
+func (P *Parser) ParseIdent() {
+	P.Trace("Ident");
+	P.Expect(Scanner.IDENT);
+}
+
+
+func (P *Parser) ParseIdentList() {
+	P.Trace("IdentList");
+	P.ParseIdent();
+	for P.tok == Scanner.COMMA {
+		P.Next();
+		P.ParseIdent();
+	}
+}
+
+
+func (P *Parser) ParseQualifiedIdent() {
+	P.Trace("QualifiedIdent");
+	P.ParseIdent();
+	if P.tok == Scanner.PERIOD {
+		P.Next();
+		P.ParseIdent();
+	}
+}
+
+
+func (P *Parser) ParseTypeName() {
+	P.Trace("TypeName");
+	P.ParseQualifiedIdent();
+}
+
+
+func (P *Parser) ParseArrayType() {
+	P.Trace("ArrayType");
+	P.Expect(Scanner.LBRACK);
+	if P.tok != Scanner.RBRACK {
+		P.ParseExpression();
+	}
+	P.Expect(Scanner.RBRACK);
+	P.ParseType();
+}
+
+
+func (P *Parser) ParseChannelType() {
+	P.Trace("ChannelType");
+	panic "ChannelType"
+}
+
+
+func (P *Parser) ParseInterfaceType() {
+	P.Trace("InterfaceType");
+	panic "InterfaceType"
+}
+
+
+func (P *Parser) ParseFunctionType() {
+	P.Trace("FunctionType");
+	panic "FunctionType"
+}
+
+
+func (P *Parser) ParseMapType() {
+	P.Trace("MapType");
+	P.Expect(Scanner.MAP);
+	P.Expect(Scanner.LBRACK);
+	P.ParseType();
+	P.Expect(Scanner.RBRACK);
+	P.ParseType();
+}
+
+
+func (P *Parser) ParseFieldDecl() {
+	P.Trace("FieldDecl");
+	P.ParseIdentList();
+	P.ParseType();
+}
+
+
+func (P *Parser) ParseStructType() {
+	P.Trace("StructType");
+	P.Expect(Scanner.STRUCT);
+	P.Expect(Scanner.LBRACE);
+	if P.tok != Scanner.RBRACE {
+		P.ParseFieldDecl();
+		for P.tok == Scanner.SEMICOLON {
+			P.Next();
+			P.ParseFieldDecl();
+		}
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	}
+	P.Expect(Scanner.RBRACE);
+}
+
+
+func (P *Parser) ParsePointerType() {
+	P.Trace("PointerType");
+	P.Expect(Scanner.MUL);
+	P.ParseType();
+}
+
+
+func (P *Parser) ParseType() {
+	P.Trace("Type");
+	switch P.tok {
+	case Scanner.IDENT:
+		P.ParseTypeName();
+	case Scanner.LBRACK:
+		P.ParseArrayType();
+	case Scanner.CHAN:
+		P.ParseChannelType();
+	case Scanner.INTERFACE:
+		P.ParseInterfaceType();
+	case Scanner.FUNC:
+		P.ParseFunctionType();
+	case Scanner.MAP:
+		P.ParseMapType();
+	case Scanner.STRUCT:
+		P.ParseStructType();
+	case Scanner.MUL:
+		P.ParsePointerType();
+	default:
+		P.Error("type expected");
+	}
+}
+
+
+func (P *Parser) ParseImportSpec() {
+	P.Trace("ImportSpec");
+	if P.tok == Scanner.PERIOD {
+		P.Next();
+	} else if P.tok == Scanner.IDENT {
+		P.Next();
+	}
+	P.Expect(Scanner.STRING);
+}
+
+
+func (P *Parser) ParseImportDecl() {
+	P.Trace("ImportDecl");
+	P.Expect(Scanner.IMPORT);
+	if P.tok == Scanner.LPAREN {
+		P.ParseImportSpec();
+		for P.tok == Scanner.SEMICOLON {
+			P.Next();
+			P.ParseImportSpec();
+		}
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	} else {
+		P.ParseImportSpec();
+	}
+}
+
+
+func (P *Parser) ParseExpressionList() {
+  P.Trace("ExpressionList");
+	P.ParseExpression();
+	for P.tok == Scanner.COMMA {
+		P.Next();
+		P.ParseExpression();
+	}
+}
+
+
+func (P *Parser) ParseConstSpec() {
+	P.Trace("ConstSpec");
+	P.ParseIdent();
+	// TODO factor this code
+	switch P.tok {
+	case Scanner.IDENT, Scanner.LBRACK, Scanner.CHAN, Scanner.INTERFACE,
+		Scanner.FUNC, Scanner.MAP, Scanner.STRUCT, Scanner.MUL:
+		P.ParseType();
+	default:
+		break;
+	}
+	if P.tok == Scanner.ASSIGN {
+		P.Next();
+		P.ParseExpression();
+	}
+}
+
+
+func (P *Parser) ParseConstDecl() {
+	P.Trace("ConstDecl");
+	P.Expect(Scanner.CONST);
+	if P.tok == Scanner.LPAREN {
+		P.ParseConstSpec();
+		for P.tok == Scanner.SEMICOLON {
+			P.Next();
+			P.ParseConstSpec();
+		}
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	} else {
+		P.ParseConstSpec();
+	}
+}
+
+
+func (P *Parser) ParseTypeSpec() {
+	P.Trace("TypeSpec");
+	P.ParseIdent();
+	// TODO factor this code
+	switch P.tok {
+	case Scanner.IDENT, Scanner.LBRACK, Scanner.CHAN, Scanner.INTERFACE,
+		Scanner.FUNC, Scanner.MAP, Scanner.STRUCT, Scanner.MUL:
+		P.ParseType();
+	default:
+		break;
+	}
+}
+
+
+func (P *Parser) ParseTypeDecl() {
+	P.Trace("TypeDecl");
+	P.Expect(Scanner.TYPE);
+	if P.tok == Scanner.LPAREN {
+		P.ParseTypeSpec();
+		for P.tok == Scanner.SEMICOLON {
+			P.Next();
+			P.ParseTypeSpec();
+		}
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	} else {
+		P.ParseTypeSpec();
+	}
+}
+
+
+func (P *Parser) ParseVarSpec() {
+	P.Trace("VarSpec");
+	P.ParseIdentList();
+	if P.tok == Scanner.ASSIGN {
+		P.Next();
+		P.ParseExpressionList();
+	} else {
+		P.ParseType();
+		if P.tok == Scanner.ASSIGN {
+			P.Next();
+			P.ParseExpressionList();
+		}
+	}
+}
+
+
+func (P *Parser) ParseVarDecl() {
+	P.Trace("VarDecl");
+	P.Expect(Scanner.VAR);
+	if P.tok == Scanner.LPAREN {
+		P.ParseVarSpec();
+		for P.tok == Scanner.SEMICOLON {
+			P.Next();
+			P.ParseVarSpec();
+		}
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	} else {
+		P.ParseVarSpec();
+	}
+}
+
+
+func (P *Parser) ParseParameterSection() {
+	P.Trace("ParameterSection");
+	P.ParseIdentList();
+	P.ParseType();
+}
+
+
+func (P *Parser) ParseParameterList() {
+	P.Trace("ParameterList");
+	P.ParseParameterSection();
+	for P.tok == Scanner.COMMA {
+		P.Next();
+		P.ParseParameterSection();
+	}
+}
+
+
+func (P *Parser) ParseParameters() {
+	P.Trace("Parameters");
+	P.Expect(Scanner.LPAREN);
+	if P.tok != Scanner.RPAREN {
+		P.ParseParameterList();
+	}
+	P.Expect(Scanner.RPAREN);
+}
+
+
+func (P *Parser) ParseResult() {
+	P.Trace("Result");
+	if P.tok == Scanner.LPAREN {
+		// TODO: here we allow empty returns - should proably fix this
+		P.ParseParameters();
+	} else {
+		P.ParseType();
+	}
+}
+
+
+// Named signatures
+//
+//        name (params)
+//        name (params) type
+//        name (params) (results)
+// (recv) name (params)
+// (recv) name (params) type
+// (recv) name (params) (results)
+
+func (P *Parser) ParseNamedSignature() {
+	P.Trace("NamedSignature");
+	if P.tok == Scanner.LPAREN {
+		P.ParseParameters();
+	}
+
+	P.ParseIdent();  // function name
+
+	P.ParseParameters();
+
+	// TODO factor this code
+	switch P.tok {
+	case Scanner.IDENT, Scanner.LBRACK, Scanner.CHAN, Scanner.INTERFACE,
+		Scanner.FUNC, Scanner.MAP, Scanner.STRUCT, Scanner.MUL, Scanner.LPAREN:
+		P.ParseResult();
+	default:
+		break;
+	}
+}
+
+
+func (P *Parser) ParseDeclaration();
+func (P *Parser) ParseStatement();
+func (P *Parser) ParseBlock();
+
+
+func (P *Parser) ParsePrimaryExprList() {
+	P.Trace("PrimaryExprList");
+	panic "PrimaryExprList"
+}
+
+
+func (P *Parser) ParseSimpleStat() {
+	P.Trace("SimpleStat");
+	P.ParseExpression();
+	switch P.tok {
+	case Scanner.ASSIGN:
+		P.Next();
+		P.ParseExpression();
+	case Scanner.COMMA:
+		P.Next();
+		P.ParsePrimaryExprList();
+	case Scanner.INC:
+		P.Next();
+	case Scanner.DEC:
+		P.Next();
+	}
+}
+
+
+func (P *Parser) ParseIfStat() {
+	P.Trace("IfStat");
+	P.Expect(Scanner.IF);
+	if P.tok != Scanner.LBRACE {
+		P.ParseSimpleStat();
+		if P.tok == Scanner.SEMICOLON {
+			P.ParseExpression();
+		}
+	}
+	P.ParseBlock();
+	if P.tok == Scanner.ELSE {
+		P.Next();
+		if P.tok == Scanner.IF {
+			P.ParseIfStat();
+		} else {
+			// TODO should be P.ParseBlock()
+			P.ParseStatement();
+		}
+	}
+}
+
+
+func (P *Parser) ParseForStat() {
+	P.Trace("ForStat");
+	panic "for stat";
+}
+
+
+func (P *Parser) ParseSwitchStat() {
+	P.Trace("SwitchStat");
+	panic "switch stat";
+}
+
+
+func (P *Parser) ParseStatement() {
+	P.Trace("Statement");
+	switch P.tok {
+	case Scanner.CONST: fallthrough;
+	case Scanner.TYPE: fallthrough;
+	case Scanner.VAR: fallthrough;
+	case Scanner.FUNC:
+		P.ParseDeclaration();
+	case Scanner.IDENT:
+		P.ParseSimpleStat();
+	case Scanner.GO:
+		panic "go statement";
+	case Scanner.RETURN:
+		panic "return statement";
+	case Scanner.BREAK:
+		panic "break statement";
+	case Scanner.CONTINUE:
+		panic "continue statement";
+	case Scanner.GOTO:
+		panic "goto statement";
+	case Scanner.LBRACE:
+		P.ParseBlock();
+	case Scanner.IF:
+		P.ParseIfStat();
+	case Scanner.FOR:
+		P.ParseForStat();
+	case Scanner.SWITCH:
+		P.ParseSwitchStat();
+	case Scanner.RANGE:
+		panic "range statement";
+	case Scanner.SELECT:
+		panic "select statement";
+	default:
+		P.Error("statement expected");
+	}
+}
+
+
+func (P *Parser) ParseStatementList() {
+	P.Trace("StatementList");
+	P.ParseStatement();
+	for P.tok == Scanner.SEMICOLON {
+		P.Next();
+		P.ParseStatement();
+	}
+}
+
+
+func (P *Parser) ParseBlock() {
+	P.Trace("Block");
+	P.Expect(Scanner.LBRACE);
+	if P.tok != Scanner.RBRACE && P.tok != Scanner.SEMICOLON {
+		P.ParseStatementList();
+	}
+	if P.tok == Scanner.SEMICOLON {
+		P.Next();
+	}
+	P.Expect(Scanner.RBRACE);
+}
+
+
+func (P *Parser) ParseFuncDecl() {
+	P.Trace("FuncDecl");
+	P.Expect(Scanner.FUNC);
+	P.ParseNamedSignature();
+	if P.tok == Scanner.SEMICOLON {
+		// forward declaration
+		P.Next();
+	} else {
+		P.ParseBlock();
+	}
+}
+
+
+func (P *Parser) ParseExportDecl() {
+	P.Trace("ExportDecl");
+	P.Next();
+}
+
+
+func (P *Parser) ParseDeclaration() {
+	P.Trace("Declaration");
+	switch P.tok {
+	case Scanner.CONST:
+		P.ParseConstDecl();
+	case Scanner.TYPE:
+		P.ParseTypeDecl();
+	case Scanner.VAR:
+		P.ParseVarDecl();
+	case Scanner.FUNC:
+		P.ParseFuncDecl();
+	case Scanner.EXPORT:
+		P.ParseExportDecl();
+	default:
+		P.Error("declaration expected");
+		P.Next();  // make progress
+	}
+}
+
+
+func (P *Parser) ParseOperand() {
+	P.Trace("Operand");
+	P.Next();
+}
+
+
+func (P *Parser) ParseSelectorOrTypeAssertion() {
+	P.Trace("SelectorOrTypeAssertion");
+}
+
+
+func (P *Parser) ParseIndexOrSlice() {
+	P.Trace("IndexOrSlice");
+}
+
+
+func (P *Parser) ParseInvocation() {
+	P.Trace("Invocation");
+}
+
+
+func (P *Parser) ParsePrimaryExpr() {
+	P.Trace("PrimaryExpr");
+	P.ParseOperand();
+	for {
+		switch P.tok {
+		case Scanner.PERIOD:
+			P.ParseSelectorOrTypeAssertion();
+		case Scanner.LBRACK:
+			P.ParseIndexOrSlice();
+		case Scanner.LPAREN:
+			P.ParseInvocation();
+		default:
+			return;
+		}
+	}
+}
+
+
+func (P *Parser) ParseUnaryExpr() {
+	P.Trace("UnaryExpr");
+	switch P.tok {
+	case Scanner.ADD: fallthrough;
+	case Scanner.SUB: fallthrough;
+	case Scanner.NOT: fallthrough;
+	case Scanner.XOR: fallthrough;
+	case Scanner.LSS: fallthrough;
+	case Scanner.GTR: fallthrough;
+	case Scanner.MUL: fallthrough;
+	case Scanner.AND:
+		P.ParseUnaryExpr();
+		return;
+	}
+	P.ParsePrimaryExpr();
+}
+
+
+func (P *Parser) ParseMultiplicativeExpr() {
+	P.Trace("MultiplicativeExpr");
+	P.ParseUnaryExpr();
+	for {
+		switch P.tok {
+		case Scanner.MUL: fallthrough;
+		case Scanner.QUO: fallthrough;
+		case Scanner.REM: fallthrough;
+		case Scanner.SHL: fallthrough;
+		case Scanner.SHR: fallthrough;
+		case Scanner.AND:
+			P.ParseUnaryExpr();
+		default:
+			return;
+		}
+	}
+}
+
+
+func (P *Parser) ParseAdditiveExpr() {
+	P.Trace("AdditiveExpr");
+	P.ParseMultiplicativeExpr();
+	for {
+		switch P.tok {
+		case Scanner.ADD: fallthrough;
+		case Scanner.SUB: fallthrough;
+		case Scanner.OR: fallthrough;
+		case Scanner.XOR:
+			P.ParseMultiplicativeExpr();
+		default:
+			return;
+		}
+	}
+}
+
+
+func (P *Parser) ParseRelationalExpr() {
+	P.Trace("RelationalExpr");
+	P.ParseAdditiveExpr();
+	switch P.tok {
+	case Scanner.EQL: fallthrough;
+	case Scanner.NEQ: fallthrough;
+	case Scanner.LSS: fallthrough;
+	case Scanner.LEQ: fallthrough;
+	case Scanner.GTR: fallthrough;
+	case Scanner.GEQ:
+		P.ParseAdditiveExpr();
+	}
+}
+
+
+func (P *Parser) ParseLANDExpr() {
+	P.Trace("LANDExpr");
+	P.ParseRelationalExpr();
+	for P.tok == Scanner.CAND {
+		P.Next();
+		P.ParseRelationalExpr();
+	}
+}
+
+
+func (P *Parser) ParseLORExpr() {
+	P.Trace("LORExpr");
+	P.ParseLANDExpr();
+	for P.tok == Scanner.COR {
+		P.Next();
+		P.ParseLANDExpr();
+	}
+}
+
+
+func (P *Parser) ParseExpression() {
+	P.Trace("Expression");
+	P.Next();
+}
+
+
+func (P *Parser) ParseProgram() {
+	P.Trace("Program");
+	P.Expect(Scanner.PACKAGE);
+	P.ParseIdent();
+	for P.tok == Scanner.IMPORT {
+		P.ParseImportDecl();
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	}
+	for P.tok != Scanner.EOF {
+		P.ParseDeclaration();
+		if P.tok == Scanner.SEMICOLON {
+			P.Next();
+		}
+	}
+}

usr/gri/src/scanner.go

@@ -4,7 +4,24 @@
 
 package Scanner
 
-export EOF;
+export
+	ILLEGAL, EOF, IDENT, STRING, NUMBER,
+	COMMA, COLON, SEMICOLON, PERIOD,
+	LPAREN, RPAREN, LBRACK, RBRACK, LBRACE, RBRACE,
+	ASSIGN, DEFINE,
+	INC, DEC, NOT,
+	AND, OR, XOR,
+	ADD, SUB, MUL, QUO, REM,
+	EQL, NEQ, LSS, LEQ, GTR, GEQ,
+	SHL, SHR,
+	ADD_ASSIGN, SUB_ASSIGN, MUL_ASSIGN, QUO_ASSIGN, REM_ASSIGN,
+	AND_ASSIGN, OR_ASSIGN, XOR_ASSIGN, SHL_ASSIGN, SHR_ASSIGN,
+	CAND, COR,
+	BREAK, CASE, CHAN, CONST, CONTINUE, DEFAULT, ELSE, EXPORT, FALLTHROUGH, FALSE,
+	FOR, FUNC, GO, GOTO, IF, IMPORT, INTERFACE, MAP, NEW, NIL, PACKAGE, RANGE,
+	RETURN, SELECT, STRUCT, SWITCH, TRUE, TYPE, VAR
+	
+	
 const (
 	ILLEGAL = iota;
 	EOF;
@@ -71,6 +88,7 @@ const (
 	KEYWORDS_BEG;
 	BREAK;
 	CASE;
+	CHAN;
 	CONST;
 	CONTINUE;
 	DEFAULT;
@@ -170,6 +188,7 @@ func TokenName(tok int) string {
 
 	case BREAK: return "break";
 	case CASE: return "case";
+	case CHAN: return "chan";
 	case CONST: return "const";
 	case CONTINUE: return "continue";
 	case DEFAULT: return "default";
@@ -234,6 +253,7 @@ type Scanner struct {
 }
 
 
+/*
 export Token
 type Token struct {
 	val int;
@@ -245,6 +265,7 @@ type Token struct {
 func (T *Token) Print () {
 	print TokenName(T.val), " [", T.beg, ", ", T.end, "[ ", T.txt, "\n";
 }
+*/
 
 
 // Read the next Unicode char into S.ch.
@@ -601,12 +622,12 @@ func (S *Scanner) Select4 (tok0, tok1, ch2, tok2, tok3 int) int {
 }
 
 
-func (S *Scanner) Scan (t *Token) (tok, beg, end int) {
+func (S *Scanner) Scan () (tok, beg, end int) {
 	S.SkipWhitespace();
 	
-	var tok int = ILLEGAL;
-	var beg int = S.pos - 1;
-	var end int = beg;
+	tok = ILLEGAL;
+	beg = S.pos - 1;
+	end = beg;
 	
 	ch := S.ch;
 	switch {
@@ -641,7 +662,7 @@ func (S *Scanner) Scan (t *Token) (tok, beg, end int) {
 			if S.ch == '/' || S.ch == '*' {
 				S.SkipComment();
 				// cannot simply return because of 6g bug
-				tok, beg, end = S.Scan(t);
+				tok, beg, end = S.Scan();
 				return tok, beg, end;
 			}
 			tok = S.Select2(QUO, QUO_ASSIGN);
@@ -659,10 +680,12 @@ func (S *Scanner) Scan (t *Token) (tok, beg, end int) {
 	
 	end = S.pos - 1;
 	
+	/*
 	t.val = tok;
 	t.beg = beg;
 	t.end = end;
 	t.txt = S.src[beg : end];
+	*/
 	
 	return tok, beg, end;
 }

usr/gri/src/test_parser.go

@@ -0,0 +1,40 @@
+// Copyright 2009 The Go Authors.  All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package main
+
+import Scanner "scanner"
+import Parser "parser"
+
+
+func Parse(src string, verbose bool) {
+	S := new(Scanner.Scanner);
+	S.Open(src);
+	
+	P := new(Parser.Parser);
+	P.Open(S, verbose);
+	
+	P.ParseProgram();
+}
+
+
+func main() {
+	verbose := false;
+	for i := 1; i < sys.argc(); i++ {
+		if sys.argv(i) == "-v" {
+			verbose = true;
+			continue;
+		}
+		
+		var src string;
+		var ok bool;
+		src, ok = sys.readfile(sys.argv(i));
+		if ok {
+			print "parsing " + sys.argv(i) + "\n";
+			Parse(src, verbose);
+		} else {
+			print "error: cannot read " + sys.argv(i) + "\n";
+		}
+	}
+}

usr/gri/src/test_scanner.go

@@ -11,9 +11,9 @@ func Scan(src string) {
 	S := new(Scanner.Scanner);
 	S.Open(src);
 	for {
-		var t Scanner.Token;
+		//var t Scanner.Token;
 		var tok, beg, end int;
-		tok, beg, end = S.Scan(&t);
+		tok, beg, end = S.Scan(/*&t*/);
 		//t.Print();  // TODO  this doesn't compile?
 		print Scanner.TokenName(tok), "\t ", src[beg : end], "\n";
 		if tok == Scanner.EOF {